ReZero: Enhancing LLM search ability by trying one-more-time

Alan Dao, Thinh Le

Retrieval-Augmented Generation (RAG) improves Large Language Model (LLM) performance on knowledge-intensive tasks but depends heavily on initial search query quality. Current metho…